Yeah, I remember reading that now.
I've read similar on people doing space sims or something, with planet positioning issues. Probably a thread on RSC about it when this was discussed before.
I'm sure though, if Ruud wanted, could write something that at least added random 'noise' into the system, something to take the blockiness out of it. Interpolating between values? Smoothing? I don't know.
I've experienced this kind of thing before with coding on my own stuff, and adding some noise in or checking other values around it would be reasonable. It's obviously costing some processing time but it makes what is essentially useless, useful.
In my case I was logging vehicle speed vs time. I used power = mass x acceleration x speed to get an instantaneous net wheel power plot. Problem was speed was logged in whole 1km/h increments, so as you can imagine my plot looked rather poor as the sample rate was quite high... even if I were accelerating at a reasonable rate, over 10 samples per second, I'd need to accelerate 10km/h per second to get a change in speed every sample, to even get a gradient to work off.
So my plot was made up of empty power values then sudden ones, then none, then sudden ones again.
I could either log rpm (much more refined, but have to hardcode and specify a gear), or look at the curve and lag the output by 1s and interpolate some new 'smoothed' values.
There will be a solution, and it's probably only needed for internal car cameras/SMD cameras, and it won't be ideal, but it at least might make the feature useful rather than just useless and annoying!
Dave